Design and Implementation of a Research Data Chatbot Pilot System Based on sLLM - Focusing on the Pilot Implementation of Llama 3.1 Korean Localization -

This study focuses on the design and implementation of a research data chatbot pilot system using a small Language Model (sLLM) and highlights a case study of Korean localization with Llama 3.1. The proposed system leverages sLLM’s efficient processing capabilities and superior natural language understanding to automate the analysis of large-scale research data and the generation of policy proposals. By integrating the advanced features of Llama 3.1, including long-context processing and enhanced tool usage, the system is tailored for complex document analysis, particularly in the Korean language context. The system’s architecture is modular, enabling scalability and adaptation to various research data environments. Experimental evaluations demonstrate the system’s superiority over traditional large language models, particularly in processing speed, summarization accuracy, and user satisfaction. The findings suggest that the system not only enhances research efficiency but also provides a secure, independent platform for data-driven policy decision-making, with potential for further optimization through real-time data processing and multi-language support.